www.nature.com/scientificreports
OPEN
received: 19 October 2015 accepted: 22 April 2016 Published: 10 May 2016
Taste substance binding elicits conformational change of taste receptor T1r heterodimer extracellular domains Eriko Nango1, Shuji Akiyama2,3, Saori Maki-Yonekura1, Yuji Ashikawa1,†, Yuko Kusakabe4, Elena Krayukhina5, Takahiro Maruno5, Susumu Uchiyama5,6, Nipawan Nuemket1,7,‡, Koji Yonekura1, Madoka Shimizu4, Nanako Atsumi7, Norihisa Yasui7, Takaaki Hikima1, Masaki Yamamoto1, Yuji Kobayashi5 & Atsuko Yamashita1,7 Sweet and umami tastes are perceived by T1r taste receptors in oral cavity. T1rs are class C G-protein coupled receptors (GPCRs), and the extracellular ligand binding domains (LBDs) of T1r1/T1r3 and T1r2/T1r3 heterodimers are responsible for binding of chemical substances eliciting umami or sweet taste. However, molecular analyses of T1r have been hampered due to the difficulties in recombinant expression and protein purification, and thus little is known about mechanisms for taste perception. Here we show the first molecular view of reception of a taste substance by a taste receptor, where the binding of the taste substance elicits a different conformational state of T1r2/T1r3 LBD heterodimer. Electron microscopy has showed a characteristic dimeric structure. Förster resonance energy transfer and X-ray solution scattering have revealed the transition of the dimerization manner of the ligand binding domains, from a widely spread to compactly organized state upon taste substance binding, which may correspond to distinct receptor functional states. Taste sensation is evoked by specific interactions between taste substances and taste receptors residing in the plasma membranes of the taste cells in taste buds in the oral cavity1,2. One of these receptors is the taste receptor type 1, the T1r family, which is evolutionarily conserved in vertebrates, including fishes, birds, and mammals3. The heterodimer of T1r2 and T1r3 recognizes sweet taste substances such as sugars and artificial sweeteners, while the heterodimer of T1r1 and T1r3 recognizes umami taste substances such as l-glutamate4–6. The T1r family proteins belong to the class C G-protein coupled receptor (GPCR) family7,8. The class C GPCR members function as constitutive homo- or heterodimers in the physiological state. The class C GPCR structure is characterized by the presence of a large extracellular domain upstream of the hepta-helical transmembrane region, which is commonly found among GPCRs. The extracellular domain consists of the ligand binding domain (LBD), responsible for primary agonist binding, followed by the cysteine rich domain (CRD), which mainly serves as a linker between the LBD and the transmembrane region (Fig. 1a). Ligand binding at the extracellular domain results in receptor activation and signal transmission to the heterotrimeric G-protein in the cytosol7,8. The receptor activation mechanism of the class A GPCR members, consisting solely of the transmembrane region, has been considered to occur via agonist binding, which changes the conformational dynamics of the protein by lowering the transition energy between the different states, and results in the transition towards the active-state 1
RIKEN SPring-8 Center, 1-1-1, Kouto, Sayo, Hyogo, 679-5148, Japan. 2Research Center of Integrative Molecular System (CIMoS), Institute for Molecular Science, National Institute of Natural Sciences, 38 Nishigo-Naka, Myodaiji, Okazaki, Aichi, 444-8585, Japan. 3Department of Functional Molecular Science, The Graduate University for Advanced Studies (SOKENDAI), 38 Nishigo-Naka, Myodaiji, Okazaki 444-8585, Japan. 4Food Research Institute, NARO, 2-1-12, Kannondai, Tsukuba, Ibaraki, 305-8642, Japan. 5Graduate School of Engineering, Osaka University, Suita, Osaka, 565-0871, Japan. 6Okazaki Institute for Integrative Biosciences, Okazaki, Aichi 444-8787, Japan. 7 Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, 1-1-1, Tsushimanaka, Kita-ku, Okayama, 700-8530, Japan. †Present address: Administration and Technology Management Center for Science and Engineering, Waseda University, Tokyo, 169-8555, Japan. ‡Present address: Japan Synchrotron Radiation Research Institute, Sayo, Hyogo, 679-5198, Japan. Correspondence and requests for materials should be addressed to A.Y. (email:
[email protected]) Scientific Reports | 6:25745 | DOI: 10.1038/srep25745
1
www.nature.com/scientificreports/
Figure 1. Taste Receptor T1r Proteins from Medaka Fish (mf). (a) Schematic drawing of the overall architecture of class C GPCR, where the codebook vector of each domain in LBD (gray dot) and the protomer torsion angle (the arrow) were depicted. (b) FSEC analysis of mf T1R2aLBD, mf T1R3LBD, and co-expression of the T1R2a and T1r3 proteins. (c) Dose-response curves for l-alanine and l-glutamine by the full-length mf T1r2a/T1r3 receptor in HEK293 cells. The error bars are ± SEM of 4–34 independent determinations. (d) The c(s) distribution of the purified mf T1r2a/3LBD, obtained from the data analysis of SV-AUC experiments.
conformation9. In contrast, the conformation of the transmembrane region of the class C GPCRs is considered to be allosterically regulated by agonist binding to the extracellular LBDs, probably through their conformational changes10–13. Accordingly, in the case of T1r, the major taste substances, including sugars and l-glutamate, are considered to target the LBD of T1r heterodimer14, and thus consequently induce the conformational change of the LBD. Due to the lack of structural information of T1r receptors, their functional mechanisms have so far been conjectured from the crystallographic observation on the other class C GPCR members. Crystal structures of the LBDs of metabotropic glutamate receptors (mGluRs) and GABAB receptor (GABABR) revealed the bilobal architecture; namely, the Venus-flytrap domain (VFTD), in which an agonist binds to the cleft between the two lobes, LB1 and LB215,16. Upon agonist binding, two types of conformational change were observed on the LBDs (Supplementary Table S1). One is the domain closure within the protomer, at the cleft between LB1 and LB2, which is referred to as the open- and closed-conformation. The other is the change between the two different forms of the dimer arrangement: the compact state with a smaller torsion angle ( ~ −50°), interpreted as the resting (R)-state (Supplementary Fig. S1)15. This view is basically compatible with the observations by Förster resonance energy transfer (FRET) of labeled mGluRs12,13. However, the actual conformations of the LBDs in the physiological state are still unknown, because available structural information is limited to those obtained in crystalline spaces. In fact, the dimer arrangements of homodimeric mGluR LBDs so far observed there were various regardless of the types of bound ligands, such as agonists, antagonists, or ligand-free15,17,18 (Supplementary Table S1), and a previous study also pointed out the possibility of biased trapping of certain conformations among dynamic conformational equilibrium by the crystal packing18. Moreover, a crystallographic analysis of the LBD of GABABR GBR1/GBR2 heterodimer revealed another mode of conformational change, where little significant dimer rearrangement is observed among apo-, antagonist bound-, and agonist bound-states16 (Supplementary Fig. S1 and Table S1). The current situation makes it obscure what kind of conformational change at the T1r LBD heterodimer is elicited by taste-substance binding. So far, not only the structural analyses but also the molecular functional analyses using the purified protein of T1rs have been completely hampered, due to the difficulties in the heterologous expression and purification even for the partial regions such as the LBD19. In this study, we found that the LBDs of T1r2 and T1r3 of medaka fish (Oryzias latipes, also known as Japanese rice fish), a representative model organism in vertebrates, can be expressed heterologously as a properly folded and functional heterodimeric protein, thus enabling various biophysical analyses. The solution-state structure analyses revealed the conformational change upon the taste substance binding to the T1r, under the condition devoid of any constraint derived from crystal packing.
Results
The ligand-binding domains of the T1r heterodimer from medaka fish exhibit proper recombinant expression. Chemosensory receptors, such as olfactory receptors and pheromone receptors, are known
to have the specific chaperone systems in their native chemosensory cells, and the proper surface expression of the receptors in heterologous cells are only achieved in the presence of the systems20–23. However, in the case for taste receptor T1rs, such specific chaperone systems are unknown20, and indeed recombinant expression of mouse- or human-T1rLBDs displayed failure of folding and localization19. Thus we first performed extensive screening of T1r genes and expression conditions to find those exhibiting proper expression, using fluorescence-detection size-exclusion chromatography (FSEC)19,24. Most of the genes from different species, as well as the numerous variations of the expression conditions including host cells, resulted in unsuccessful protein production, as reported previously19. Nevertheless, the LBDs of T1r2a and T1r3 from medaka fish (mf)25 showed good secretion and a sharp FSEC elution peak corresponding to the dimer species, indicating proper folding and oligomerization, only when they were co-expressed in insect cells (Fig. 1b). mf T1r3LBD showed fair conservation with mammalian T1r3LBD (~37% similarity). On the other hand, LBD of mf T1r2a, one of the three T1r2 subtypes in medaka fish, showed moderate conservation with both mammalian T1r1LBD and T1r2LBD (37~39% similarity), reflecting Scientific Reports | 6:25745 | DOI: 10.1038/srep25745
2
www.nature.com/scientificreports/
Figure 2. Electron microscopic observation of T1r2a/3LBD. The top panels are negative-staining raw particle images of the purified T1r2a/3LBD, with close-up views of representative particles in the insets. The bottom panels are the representative two-dimensional class averages of particles. (a) The l-glutamine-bound state. (b) The ligand-free state. the fact that the mfT1r2a/T1r3 heterodimer responds to a wide array of l-amino acids, but not to sugars or artificial sweeteners such as saccharin26. Dose-response measurement of the full-length mf T1r2a and mf T1r3 heterodimer cloned in this study confirmed the similar EC50 value for l-alanine (2.70 ± 1.20 mM) to that reported previously26, and revealed an even higher affinity to l-glutamine, with the EC50 value of 100 ± 26.0 μM (Fig. 1c and Supplementary Table S2). Because medaka fish reportedly show preferences to the foods containing amino acids27, l-glutamine and alanine are considered to serve as taste substances to medaka fish. Therefore, the mf T1r2a/mf T1r3-LBD heterodimer and its amino-acid binding are expected to serve as the first molecular platform to assess the general structural and functional properties T1r LBD heterodimer, including those for taste substance binding.
T1r2aLBD and T1r3LBD form a stable heterodimer. mf T1r2aLBD and mfT1r3LBD were successfully co-purified after recombinant expression in insect cells (Supplementary Fig. S2). The purified protein exhibited a monodisperse distribution as confirmed by a sedimentation velocity analytical ultracentrifugation (SV-AUC) analysis. In the obtained c(s) distribution, a single peak with an estimated molecular weight of 108.5 kDa was observed (Fig. 1d). This result clearly indicated that the purified T1r2aLBD and T1r3LBD (53.5 and 55.2 kDa respectively, as estimated from the amino acid sequences) exclusively formed a stable heterodimer. To visualize the structural organization of T1r2a/3LBD, we performed the electron microscopic observation of negatively stained T1r2a/3LBD (Fig. 2). The particles exhibit the presence of two segments, most likely corresponding to each protomer of T1r2a and T1r3 LBD proteins, both in the presence or absence of a taste substance l-glutamine. Two-dimensional averages of T1r2a/3LBD particle exhibited a bilobal feature, characteristic to VFTD. The approximate particle sizes were observed as ~95 × ~75 Å, which agrees with those for other class C GPCR LBD dimer structures. These observations indicated that the manner of dimerization of T1r2a/3LBD is similar to that observed on LBDs of other class C GPCRs, such as mGluRs. Conformational change of the ligand binding domains upon taste substance binding. So far, all the structural analyses of other class C GPCR LBDs were performed by crystallography, using deglycosylated samples in many cases15–18. In this study, we performed multiple structural analyses in both the presence and absence of a taste substance, using the glycosylated protein sample (Supplementary Fig. S2) in solution, which is closer to a physiological condition without any constraint derived from crystal packing. We first assessed whether the binding of a taste substance to the T1r LBD induces a conformational change of the protein. To answer this question, the T1r2a/3LBD heterodimer fused with a fluorescence protein, either
Scientific Reports | 6:25745 | DOI: 10.1038/srep25745
3
www.nature.com/scientificreports/
Figure 3. Conformational change of T1r2a/3 LBD upon taste substance binding. (a) Dose-dependent FRET signal changes of the T1r2aLBD-Cerulean and T1r3LBD-Venus heterodimer for taste substance binding. The error bars are ± SEM of 3 independent determinations. (b,c) l-Glutamine (b) and l-alanine (c) binding to mfT1r2a/3LBD, measured by isothermal titration calorimetry. The upper and lower panels show the raw data and the integrated heat signals upon ligand injection, respectively, with binding isotherms fitted assuming 1 ligand: 1 heterodimer binding.
Cerulean (a CFP variant) or Venus (a YFP variant)28, at their C-termini was subjected to Förster resonance energy transfer (FRET) analyses. In the presence of the taste substances l-glutamine and l-alanine, the labeled T1r2a/3LBD exhibited elevated FRET signals (Fig. 3a). The EC50 values for the FRET signal changes upon the l-glutamine and l-alanine titration were 12.7 ± 2.7 and 168 ± 19.0 μM, respectively (Supplementary Table S2). We confirmed that the FRET signal rise accompanies the taste substance binding, as the EC50 values for the former are close to the Kd values of l-glutamine and l-alanine to the non-labeled T1r2a/3LBD analyzed by isothermal titration calorimetry (Fig. 3b,c and Supplementary Table S2). These results indicated that taste substance binding to the T1r2a/3LBD induces a conformational change of the protein, in most likely which the C-termini of the LBDs of the two protomers come closer to each other. The order of the EC50 values of two amino acids was coincident with that for the receptor responses as observed above. Although the EC50 values for the FRET change by the labeled-LBD and the response by the full-length receptor have 8~16 fold differences, such deviations were often observed on the other receptors when the different assay conditions/methods were used, because the ligand efficacies are affected by the intrinsic receptor characteristics, the downstream signaling pathways, and so on29. For example, mGluR1 exhibited the Kd of 1.3 μM for the l-glutamate binding to the purified LBD measured by fluorescence change30, while EC50 of 4.6~22 μM for the l-glutamate response measured by Ca2+ -activated Cl− current in a Xenopus oocyte expression system (3.5~17 fold differences)31. Therefore, the results observed in this study suggested that the conformational transition is relevant to the receptor responses. To analyze the conformational changes in further detail, the non-labeled T1r2a/3LBD, in the presence or absence of l-glutamine, was subjected to small-angle X-ray scattering (SAXS) analyses (Fig. 4). The molecular mass estimated on the bases of the forward scattering (121~123 kDa) as well as the Porod volume (144~150 kDa) was nearly constant irrespective of the presence of l-glutamine (Supplementary Table S3), exhibiting a fair agreement with the sum of those for T1r2aLBD and T1r3LBD determined by mass spectroscopy and SDS-PAGE (127 kDa; Supplementary Fig. S2). The radius of gyration (Rg) in the presence of glutamine (37.0 ± 0.5 Å) was smaller than that of the ligand-free protein (39.8 ± 0.6 Å; Supplementary Table S3). This difference in the Rg values amounted to ~7.6%, a comparable value to those analyzed by the previous SAXS measurements on the nucleotide-mediated conformational change of the myosin head domains (4.5~5.9%)32 or the open-close transition of phosphoglycerate kinase during catalysis (~8.0%)33. This result clearly indicates that a significant conformational change occurs between the two states. The radical conformational switching was also supported by the pair distance distribution functions, P(r), which exhibited a ~30 Å reduction in the maximum dimension of particles (Dmax) upon glutamine binding (Fig. 4b and Supplementary Table S3). Judging from the crystal structures of other class C GPCR LBDs, the observed particle size difference seems not be solely attributed to the conformational change within a protomer, but to the rearrangement of the dimerization. For example, the open- (O) to close (C) transitions within the same dimerization state, such between PDB IDs 3KS9 to 1EWK (mGluR1) or 4MQE to 4MS3 (GABABR), results less than 3 Å differences, while the R- to A-state transitions of the dimerization state, such between 1EWT to 1EWK Scientific Reports | 6:25745 | DOI: 10.1038/srep25745
4
www.nature.com/scientificreports/
Figure 4. Overall shapes of T1r2a/3 LBD in ligand-free and l-glutamine-bound states revealed by SAXS. (a) SAXS curves of the ligand-free (red) and l-glutamine-bound (blue) forms of T1r2a/3 LBD. The inset indicates the Guinier plots of the ligand-free (red) and l-Gln-bound (blue) forms of T1r2a/3 LBD, for which the Guinier analyses were conducted by using the Q range (highlighted data points in the inset) from 0.01003 Å−1 to Qmax